I'm encountering a build problem for Python 2.6.4 on Snow Leopard.
- Mac OS X 10.6
- Yonah CPU, 32-bit
- gcc-4.2.1
Update I
Solved by removing all non-standard includes and libraries from CFLAGS (there happened to be a uuid/uuid.h
in there ...). Still, it compiled despite the error describe below, with /usr/include/hfs/hfs_format.h:765
being a hot spot. For the curious or resourceful, the source file in question:
$ cat /usr/include/hfs/hfs_format.h
...
748 #include <uuid/uuid.h>
749
750 /* JournalInfoBlock - Structure that describes where our journal lives */
751
752 // the original size of the reserved field in the JournalInfoBlock was
753 // 32*sizeof(u_int32_t). To keep the total size of the structure the
754 // same we subtract the size of new fields (currently: ext_jnl_uuid and
755 // machine_uuid). If you add additional fields, place them before the
756 // reserved field and subtract their size in this macro.
757 //
758 #define JIB_RESERVED_SIZE ((32*sizeof(u_int32_t)) - sizeof(uuid_string_t) - 48)
759
760 struct JournalInfoBlock {
761 u_int32_t flags;
762 u_int32_t device_signature[8]; // signature used to locate device.
763 u_int64_t offset; // byte offset to the journal on the device
764 u_int64_t size; // size in bytes of the journal
765 uuid_string_t ext_jnl_uuid;
766 char machine_serial_num[48];
767 char reserved[JIB_RESERVED_SIZE];
768 } __attribute__((aligned(2), packed));
769 typedef struct JournalInfoBlock JournalInfoBlock;
...
I'm leaving the qu开发者_开发技巧estion open, since the build yielded too much warnings and this error and still puzzles me a bit ...
Update II
To get rid of the warnings regarding the deployment target, I edited the Makefile
before compilation:
$ cat Makefile
...
126 MACOSX_DEPLOYMENT_TARGET=10.3 # => 10.6
...
Original Question
When trying to build Python 2.6.4 from source, I run into an error:
$ uname -v
$ Darwin Kernel Version 10.2.0: Tue Nov 3 10:37:10 PST 2009; \
root:xnu-1486.2.11~1/RELEASE_I386
$ cd ~/src/Python-2.6.4
$ ./configure --enable-universalsdk=/ --prefix=$HOME
checking for --with-universal-archs... 32-bit
...
checking machine type as reported by uname -m... i386
...
checking for OSX 10.5 SDK or later... no // <- But I have XCode + the SDKs installed?
...
$ make
...
...
/usr/include/hfs/hfs_format.h:765: error: \
expected specifier-qualifier-list before ‘uuid_string_t’
It seems to root in Python/mactoolboxglue.c
. Hints welcome!
Also I get a lot of warnings of these kinds:
/usr/include/AvailabilityMacros.h:108:14: warning: #warning Building for \
Intel with Mac OS X Deployment Target < 10.4 is invalid.
gcc -c -arch ppc -arch i386 -isysroot / -fno-strict-aliasing -DNDEBUG -g \
-fwrapv -O3 -Wall -Wstrict-prototypes -I. -IInclude -I./Include \
-DPy_BUILD_CORE -o Objects/structseq.o Objects/structseq.c
In file included from /usr/include/architecture/i386/math.h:626,
from /usr/include/math.h:28,
from Include/pyport.h:235,
from Include/Python.h:58,
from Objects/structseq.c:4:
Try adding --universal-archs=32-bit
to the configure
arguments.
EDIT: You may also need to set environment variable MACOSX_DEPLOYMENT_TARGET=10.6
and explicitly use the 10.6 SDK:
export MACOSX_DEPLOYMENT_TARGET=10.6
./configure --universal-archs=32-bit --enable-universalsdk=/Developer/SDKs/MacOSX10.6.sdk ...
There are still some configure issues with building Python on 10.6. If you re-use the build directory, make sure you clean out all of the cached files that previous runs of configure
may have left behind.
BTW, if you just need a 32-bit version, you could use the 2.6.4 OS X installer from python.org.
It seems that you aren't the only one running into this error:
Not sure if this has the same root cause :
http://trac.macports.org/ticket/21282
Scroll down to the end comments,which seem to indicate a positive outcome. (repeated here for convenience):
...//Quote from trac.macports.org:
"try to rename, move or delete /opt/local/include/uuid/uuid.h" ...//End Quote
精彩评论